Epilepsy is the most prevalent disabling neurologic illness, and depression is the most frequent comorbid condition associated with epilepsy. The prevalence of depression is 20-50 percent in patients with uncontrolled seizures. This combination affects between 250,000 and 450,000 people in the United States. Our recent clinical studies have shown that depression is a strong predictor of function and health outcomes in epilepsy. Despite the marked adverse effects and high prevalence of depression in epilepsy, most affected patients are not treated. This complacency toward treatment may result from insufficient use of diagnostic screening, the widespread belief that antidepressants lower the seizure threshold, or lack of demonstrated efficacy in the only controlled trial of antidepressant medications in epilepsy. The broad aims of this study are to define the benefits of antidepressant treatment on mood, compliance, and health outcomes in epilepsy patients with comorbid major depression. Based on our prior clinical and research experience, we hypothesize that 1) pharmacotherapy or psychotherapy will reduce depression and improve health-related quality of life in patients with refractory epilepsy, 2) antiepileptic medication compliance will improve after reduction of depression, 3) seizure frequency will not significantly increase during treatment with a selective seratonin reuptake inhibitor compared to psychotherapy, and 4) depression and antiepileptic medication toxicity are stronger predictors of health-related quality of life than seizure frequency or severity in patients with refractory epilepsy. The hypotheses will be tested through a randomized trial comparing the efficacy of sertraline (n=127) to cognitive behavior therapy (n=127) for mood and health outcomes in patients with refractory epilepsy and depression. Reliable and valid measures will be used to assess depression and health-related quality of life. Electronic, computer-assisted monitoring will determine compliance. Multivariate repeated-measures analyses will be used to determine the interrelationships of treatment, mood, antiepileptic medication toxicity, seizure frequency and severity, compliance and health-related quality of life. We anticipate that dissemination of the results of a positive study will support the modification of the current model of intervention for epilepsy from predominantly seizure reduction to a more comprehensive approach that includes assessment and treatment of depression